arguing agents in a multi- agent system for regulated information exchange pieter dijkstra

18
Arguing Agents in a Multi-Agent System for Regulated Information Exchange Pieter Dijkstra

Upload: leon-boyd

Post on 04-Jan-2016

216 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Arguing Agents in a Multi-Agent System for

Regulated Information Exchange

Pieter Dijkstra

Page 2: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Regulated information exchange Information exchange is often regulated

by data protection laws Hardcoding these laws in

communication protocols: Ensures compliance with the law But in a rigid way, ignoring exceptional

circumstances, social goals ...

Allow for argumentation

Page 3: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

ANITA: MAS for exchanging crime-related information

Goal of police organisation: exchange as much information as possible But stay within the law

Goal of crime investigators: protect their investigation Anonymity of informants!

How to balance these goals? Allow agents to argue with each other; But also to reason internally about their goals

Page 4: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Example P: Tell me all you know about recent

trading in explosive materials (request)

P: why don’t you want to tell me?

P: why aren’t you allowed to tell me?

P: You may be right in general (concede) but in this case there is an exception since this is a matter of national importance

P: since we have heard about a possible terrorist attack

P: OK, I agree (offer accepted).

O: No I won’t (reject)

O: since I am not allowed to tell you

O: since sharing such information could endanger an investigation

O: Why is this a matter of national importance?

O: I concede that there is an exception, so I retract that I am not allowed to tell you. I will tell you on the condition that you don’t exchange the information with other police officers (offer)

Page 5: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Example P: Tell me all you know about recent

trading in explosive materials (request)

P: why don’t you want to tell me?

P: why aren’t you allowed to tell me?

P: You may be right in general (concede) but in this case there is an exception since this is a matter of national importance

P: since we have heard about a possible

terrorist attack

P: OK, I agree (offer accepted).

O: No I won’t (reject)

O: since I am not allowed to tell you

O: since sharing such information could endanger an investigation

O: Why is this a matter of national importance?

O: I concede that there is an exception, so I retract that I am not allowed to tell you. I will tell you on the condition that you don’t exchange the information with other police officers (offer)

Page 6: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Example P: Tell me all you know about recent

trading in explosive materials (request)

P: why don’t you want to tell me?

P: why aren’t you allowed to tell me?

P: You may be right in general (concede) but in this case there is an exception since this is a matter of national importance

P: since we have heard about a possible

terrorist attack

P: OK, I agree (offer accepted).

O: No I won’t (reject)

O: since I am not allowed to tell you

O: since sharing such information could endanger an investigation

O: Why is this a matter of national importance?

O: I concede that there is an exception, so I retract that I am not allowed to tell you. I will tell you on the condition that you don’t exchange the information with other police officers (offer)

Page 7: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

The communication language

Speech act Attack Surrender

request() offer (’), reject() -

offer() offer(’) ( ≠ ’), reject() accept()

reject() offer(’) ( ≠ ’), why-reject ()

-

accept() - -

why-reject() claim (’) -

claim() why() concede()

why() since S (an argument) retract()

since S why() ( S)deny() ( S)’ since S’ (a defeater)

concede() concede ’ (’ S)

concede() - -

retract() - -

deny() - -

Page 8: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

The protocol Start with a request Repy to a previous move of the other agent Pick your replies from the table Finish persuasion before resuming negotiation Turntaking:

In nego: after each move In pers: various rules possible

Termination: In nego: if offer is accepted or someone withdraws In pers: if main claim is retracted or conceded

Page 9: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Example dialogue formalised

P: Request to tell

O: Reject to tell

P: Why reject to tell?

Embedded persuasion

...

O: Offer to tell if no further exchange

P: Accept after tell no further exchange

Page 10: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Persuasion part formalisedO: Claim Not allowed to tell

P: Why not allowed to tell?

O: Not allowed to tell since telling endangers investigation &What endangers an investigation is not allowed

P: Concede What endangers an investigation is not allowed

O: Why National importance?

P: National importance since Terrorist threat &Terrorist threat National importance

P: Exception to R1 since National importance & National importance Exception to R1

Page 11: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Persuasion part formalisedO: Claim Not allowed to tell

P: Why not allowed to tell?

O: Not allowed to tell since telling endangers investigation &What endangers an investigation is not allowed

P: Concede What endangers an investigation is not allowed

O: Why National importance?

P: National importance since Terrorist threat &Terrorist threat National importance

P: Exception to R1 since National importance & National importance Exception to R1

P: Concede Exception to R1

Page 12: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Persuasion part formalisedO: Claim Not allowed to tell

P: Why not allowed to tell?

O: Not allowed to tell since telling endangers investigation &What endangers an investigation is not allowed

P: Concede What endangers an investigation is not allowed

O: Why National importance?

P: National importance since Terrorist threat &Terrorist threat National importance

P: Exception to R1 since National importance & National importance Exception to R1

O: Concede Exception to R1

O: Retract Not allowed to tell

Page 13: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Agent Design Knowledge of

Regulations Goals Consequences of actions

Reasoning Defeasible

Dialogue policies Negotiation Persuasion

Belief revision policies

Page 14: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Negotiation policy of responding agent

Perform requested action?

Obliged? yes: accept no: →

Forbidden? yes: reject no: →

Violation of own interests? no: accept yes: →

Try to find conditions yes: counteroffer no: reject

Page 15: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Persuasion policy for responding agent (1)

How to respond to “p since Q”?

Does the argument satisfy the context criteria? yes: concede premises and conclusion no: →

Does KB imply p? yes: concede conclusion no: →

Does KB warrant a counterargument (for not-p or an exception)?

yes: state counterargument yes or no: →

Investigate each premise q in Q

Page 16: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Persuasion policy for responding agent (2)

How to respond to premise q of “p since Q”?

Is the argument of the form p since p? yes: deny p no: →

Does KB imply q? yes: concede q no: →

Does KB imply not-q? yes: state argument for not-q no: why q

Page 17: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Persuasion policy for responding agent (3)

How to respond to “why p”?

Does KB warrant an argument p since Q? yes: state “p since Q” no: retract p

Page 18: Arguing Agents in a Multi- Agent System for Regulated Information Exchange Pieter Dijkstra

Conclusion

We have integrated three strands of theoretical work on dialogue in a MAS application scenario: Argumentation logics Dialogue systems Dialogue strategies for agents